The Tangible Value of Experiential Learning in M&A New Evidence from Takeover of Experienced Deal-Makers

Transcription

1 The Tangible Value of Experiential Learning in M&A New Evidence from Takeover of Experienced Deal-Makers Dr. Indrajeet Mohite* Abstract Organisational learning theory predicts that firms and their top executives should get better in M&A deals with experience. Yet, existing studies on acquisition learning document mixed results and point, at best, to a negative association between deal experience and acquirer returns initially, with a partial turnaround only late in a deal sequence. The lower gains in subsequent acquisition deals are likely induced by exogenous factors, such as the serial acquirer s declining investment opportunity set, which can conceal the acquirer s potential to learn with experience. To tackle this issue this study examines the value of M&A experience by concentrating on the target firms prior acquisitiveness and investigates whether experienced deal-makers learn to negotiate the deal in favour of their shareholders when they are takeover by other firms. I find that the value created by the acquirer is inversely related to the deal-making experience of the target firm. And, the premium received by the target shareholders is positively related to the target's deal-making experience. Our findings offer valuable contributions to the M&A learning literature as they suggest that deal making skills and negotiation ability improve with experience resulting in target firms securing more benefits for their shareholders at the expense of acquirers. Keywords: M&A Learning, Acquisition experience, Mergers and Acquisitions, Serial acquisitions, Investment decisions. JEL classification: G02, G14, G34 * Indrajeet is from ICMA Centre, Henley Business School, University of Reading, Whiteknights Campus, U.K. I would like to thank Dr. George Alexandridis, Prof. Raghavendra Rau, Dr. Carol Padgett, Prof. Chris Brooks, Prof. Gulnur Muradoglu, Prof. Jerry Coakley, seminar participants at the Behavioural Finance Working Group, the ICMA Centre and the Young Financial Scholars Conference for their useful comments and suggestions. All errors are my own. 1

2 1. Introduction Corporations have become a dominant force in our society and their decisions shape all our daily lives. 1 This study investigates experiential learning in one of the most important corporate investment decisions, Mergers and Acquisitions. In recent years, the worldwide investment in M&A has reached record levels. 2 In 2013, the total global M&A volume reached $2.9 trillion. 3 M&As have, indeed, become a significant component of the world economy. However, it is rather alarming that studies on wealth created in M&As find that acquirers do not create significant value for their shareholders (Jensen and Ruback, 1983 and Andrade, Mitchell and Stafford, 2001). It has often been argued that acquirers can improve their acquisition performance as they gain deal-making experience such that acquisitions of firms with deal-making experience will outperform those without such experience (Lubatkin, 1983). Intuitively, the inferences drawn from previous deals (Haleblian and Finckelstein, 1999) and the threat of being disciplined for undertaking poor quality deals (Mitchell and Lehn, 1990 and Lehn and Zhao, 2006) should aid the acquirers to learn from their experience, and as a result undertake better deals in the future. The learning curve theory predicts that if acquirers learn by doing M&A then the acquirers acquisition performance will improve with their acquisition experience (Dutton, Thomas and Butler, 1984 and Lieberman, 1987). However, external factors can affect this relationship such as the acquirer s time-varying investment opportunity set (Klasa and Stegemoller, 1 This is best illustrated by Serafaim (2013), "In 1980 the world s largest 1,000 publically listed companies made $2.64 trillion in revenue, or $7.0 trillion in 2012 dollars, adjusted using the consumer price index. They directly employed nearly 21 million people, and had a total market capitalization of close to $900 billion ($2.4 trillion in 2012 dollars), or 33 percent of the world total. By 2012, the Global 1000 made $34 trillion in revenue. They directly employed 73 million people, hundreds of millions in their supply chains, and had a total market cap of $28 trillion. These companies and their supply chains have an enormous potential to confer both good and ill on society. They create goods and services for customers, wealth for their shareholders, and jobs for millions of people." 2 For additional information, please see the article titled "Global M&A at 7-year high as big deals return" published on 30th June, 2014 by Reuters for 3 For more details, please see Dealogic global M&A review,

3 2007). So, simply finding a relationship between acquirer gains and acquisition experience is not sufficient to test learning unless other factors that can affect this relationship are controlled for. For example, Billet and Qian (2008) find that serial acquirers higher-order deals create less value than their initial deals and conclude that acquirers do not learn, but actually become hubristic as they gain experience. However, Klasa and Stegemoller (2007) find similar result but attribute it to the changes in the firms investment opportunity set, rather than the acquirers growing hubris. They show that the acquirer gains diminish along the acquirers deal-series because of the reduction in the acquirer s investment opportunity set. Given the challenges to assess the effects of experience, certainly, serial acquirers might have the potential to learn from their past experience and create value in their subsequent deals, but in order to accurately test the value of deal-making experience in M&A; we need an appropriate platform that is free from conflicting effects of other external effects. In this study, I provide this relevant platform to examine whether corporate managers learn from past M&As. I examine how experienced acquiring managers apply their deal-making experience when their firm is acquired by another firm. Surprisingly, this has not been examined by previous studies on experiential learning in M&A. Analysing the targeted acquirer s acquisition experience, in contrast to that of the acquirer s, can offer a robust test of whether deal-making experience can actually create value in M&A because such analysis is based on a unique sample (i.e. a sample of experienced target firms) that is not affected by the conflicting effects of diminishing returns that affects acquirers CARs along the dealorder. Furthermore, Fuller, Netter and Stegemoller (2002) show that the gains to the acquirer depend on the bargaining ability of the target. They argue that one of the main reasons behind negative announcement returns to the acquiring shareholders when acquiring public targets, is the strong negotiating position held by public targets. However, it remains unclear how public targets become good negotiators in acquisitions. They could naturally be skilled 3

4 negotiators or alternately, they could develop this skill from their past acquisition experience. According to the 2013 M&A outlook report, 33% of corporates mentioned 'Deal Term Negotiation' as the most difficult aspect of M&A. 4 Even the success of 2013's largest deal - Vodafone's sale of its stake in the Verizon Wireless joint venture for $ 130 billion was attributed to the deal making experience of the two groups 5. The impact of experience on negotiation performance was demonstrated experimentally by Thompson (1990). In her seminal article, Thompson (1990) shows that negotiation performance improves with experience and most importantly negotiators were able to apply the negotiation skills learned in one bargaining task to their other negotiation decisions. She finds that the improvement in performance with experience comes from greater judgement accuracy about the opponent and higher bargaining aspiration. Moreover, looking at the ability to learn negotiation skills, Boven, Nadler and Thompson (2003) show that accumulation of tacit knowledge through previous observations can increase the negotiation performance. Additionally, looking at the ability to learn in M&A, Haleblian and Fincklestein (1999) suggest that acquirers can draw inferences from their past acquisitions and apply these inferences to improve the quality of their subsequent deals. But, the question remains, can acquirers apply these inferences when they get acquired by another firm? The above findings suggest that the target firm can draw inferences from its previous acquisitions and apply them to negotiate a better deal for their shareholders, thus retaining a larger portion of the gain from the deal, which will consequently result in the reduction of the gain available from the deal to their acquirer. I investigate if targets learn from their previous acquisition experience. Specifically, if targets learn from their previous acquisition experience, then we should see following characteristics when it is acquired. First, acquirer gains will reduce when targets have more acquisition 4 Source: M&A Outlook 2013, Mergermarket and R.R. Donnelly. 5 Source: Capital Insights from EY Transaction Advisory Services. 4

5 experience. And, this should be further supported by the second characteristic; the premium received by the target should rise with increase in target's acquisition experience. I find evidence supporting the above conjectures. In this study, I use a sample of U.S. public acquisition during Over this period, U.S. public companies acquired $ 4.38 trillion (in constant year 2010 dollars) worth of other U.S. public companies. I begin by examining the acquirer cumulative abnormal returns (CARs) over the 3-day (-1,+1) event window around the announcement date classified by target's acquisition experience. I follow a conservative approach in defining target's acquisition experience. It is defined as the number of deals completed by the target in the prior 10 years or at least prior 5 years when the acquisition history of prior 10 years in not available. I obtain similar results when I measure the acquisition experience over the entire sample period and when I measure the recent acquisition experience in the prior 5 year period. Although the choice of 10 years is arbitrary, Billet and Qian (2008) argue that the duration of at least preceding 5 years allows sufficient time span for the acquisition history to develop and it is also short enough so that previous acquisitions are likely to be informative while drawing inferences. 6 The mean acquirer abnormal return when targets do not have any acquisition experience is -1.16%, while it is -2.95% when targets have acquisition experience of at least 5 deals. The difference of percentage points is significant at 1% level and reveals the negative relation between acquirer CARs and the targets acquisition experience. In order to isolate and capture the clean effect of the target's acquisition experience on the acquirer abnormal returns, I calculate adjusted acquirer CAR. Adjusted acquirer CAR is the difference between acquirer CAR to the acquirer of the target with acquisition experience and the median acquirer CAR to the acquirers of similar targets with no acquisition experience. 6 The acquisition history consists of deals that are economically significant with no restriction on the target listing status. The deal is considered to be economically significant if the relative size of the deal to acquirer's pre-deal market capitalisation is at least 1% and the transaction value is at least $ 1 million (Moeller, Schlingemann and Stulz, 2004). 5

6 Since I match the deal and the target characteristics, it allows me to compare acquirer CARs of similar targets that were acquired during similar period, but most importantly these targets have different levels of acquisition experience. 7 Hence, the difference in acquirer CARs (i.e. adjusted acquirer CAR) is more likely to reflect the isolated effect of target's acquisition experience on acquirer CAR. 8 The mean adjusted acquirer CAR when targets have acquisition experience of not more than 1 deal is -0.03%, while it is -2.14% when targets have acquisition experience of at least 5 deals. The difference of percentage points is significant at 1% level and shows that adjusted acquirer CARs are negatively related to the targets acquisition experience. Next, I control for other determinant of acquirer CARs. In the multivariate framework I control for deal and firm characteristics along with year and industry effects and find consistent evidence. Acquirer CARs and adjusted acquirer CARs are significantly negatively related to the targets' acquisition experience. Specifically, just 1 deal increase in the targets acquisition experience reduces the gains to the acquiring shareholders by 0.16 percentage points, a decline of 11.35% given the mean acquirer CAR of -1.41%. 9 If we consider the mean value of the acquirer market capitalisation ($ mil) this translates into a reduction of $ 17.4 million in value for the acquiring shareholders over a 3 day period around the announcement date. Then, I examine the premium offered to the target shareholders. I find that premium and adjusted premium are significantly positively related to the targets acquisition experience. 10 Specifically, just 1 deal increase in the targets acquisition 7 Adjusted acquirer CAR (-1,+1) is calculated as acquirer CAR minus the median acquirer CAR of control sample of deals that were completed in the same year or one year before or one year after and involved target in the same industry, same listing status, similar relative size (+/- 10%) and similar target MTB (+/- 10%). The control sample of deals includes acquisitions of targets without acquisition experience. For matching the deals, target industry is classified according to Fama and French 12 industry classification. 8 Adjusted acquirer CAR also controls for the changing opportunity set of the acquirer (Klasa and Stegemoller, 2007). If the target and the deal characteristics are a function of the acquirer's evolving opportunity set, then comparing the acquirer's acquisition performance for targets with similar firm and deal characteristics but with different level of acquisition experience will allow us to capture the genuine effect of target's acquisition experience on acquirer CARs, rather than due to acquirer's changing opportunity set. 9 From Table 6 - specification (3). 10 Adjusted premiums are calculated using the same methodology followed to calculate adjusted acquirer CARs. 6

7 experience increases the premium to the target shareholders by 0.70 percentage points. 11 If we consider the mean value of the target market capitalisation ($ mil) this equals $ 8.8 million increase in the value of the premium received by the target shareholders. Finally, I examine the acquirer's post-event abnormal stock performance. I document that the acquirer's post-event abnormal stock performance is significantly negatively related to the target s acquisition experience. This indicates that the market does not reverse its initial perception of the quality of the deal, and in fact it is likely to be an under-reaction. Moreover, acquirer's poor post-event abnormal stock performance might reflect the integration complexities experienced by the acquirer in realising the expected synergy from the deal while integrating targets with more acquisition experience, because such targets that previously acquired other businesses are likely to have a complex business structure, which can be challenging for the acquirer to integrate. Taken as a whole, the combination of findings provides support for the conceptual premise that targets with experience are able to negotiate better deals for their shareholders and reduce the gains available from the deal to the acquirer. The results are robust to alternate specifications of target's acquisition experience, acquirer CAR and premium. The evidence suggests that acquirers should be particularly careful when they are acquiring experienced deal-makers. This study has several important contributions. Firstly, it contributes to the empirical literature of behavioural finance by documenting clean evidence, for the first time, that dealmaking experience can create value in M&A after examining a unique sample of target firms unaffected by the conflicting effects of diminishing returns due to external factors. Secondly, this study demonstrates for the first time that all else equal, deal-making experience can result in target shareholders extracting more benefits from the transaction through securing a higher 11 From Table 7 - specification (3). 7

8 acquisition premium. This has significant implications for the literature on the value of managerial experience in improving negotiation skills. As a result of targets being able to negotiate better deals, acquiring firms are subject to more negative abnormal returns when they acquire targets having managers with superior skills. Hence, lastly, this study also provides new evidence on the importance of target deal-making experience as a determinant of returns to acquiring firms. It adds to the empirical literature of M&A by illustrating that the well-documented reduction in acquirer gains in public acquisitions is magnified by the deal-making experience of the target firm. The paper is organised as follows. Section 2 describes the sample selection process, the methodology followed and reports the sample statistics. Section 3 presents the empirical analysis of acquirer gains. Section 4 examines the premiums. Section 5 investigates the postevent abnormal stock performance. And, I conclude in Section Data, Methodology and Sample Description 2.1 Data This section describes the sample selection methodology. To begin with, I identify the acquisition history of all U.S. public acquirers and construct a database consisting of the deals completed by all U.S. public acquirers between 1990 and 2010, involving public, private or subsidiary target with transaction value of at least $ 1 million and relative size of the deal to the acquirer's size of at least 1%. 12 From this database of acquisitions I identify acquisitions of public targets. 13 This is the primary sample. The initial acquisition database facilitates in the identification of the acquisition history and the measurement of the acquisition experience of all public acquirers, and most importantly, of all public targets in the primary sample. 12 All dollar values are in terms of 2010 dollar values. Monetary measures are reported in inflation-adjusted 2010 dollar values using the Consumer Price Index (CPI) from the data library on the website of Robert Shiller. 13 For the acquisition history database, N = deals. For the primary sample, N = 2355 deals. 8

9 Next, in order to calculate the experience adjusted performance measures, I separate the sample into 2 parts: 1) sub-sample of targets with acquisition experience and 2) control sample of targets with no acquisition experience. I use the control sample of targets with no acquisition experience to adjust the performance of the sub-sample of targets with acquisition experience. The following section describes the sample selection process, the methodology implemented and the variable definitions. The initial acquisition history database consists of all the M&A deals from 1 st January, 1990 up to 31 st December, 2010 that were completed by a U.S. public listed firm. The acquisition data is obtained from SDC M&A database. I exclude spin-offs, recapitalizations, self-tenders, repurchases, minority stake purchases, acquisitions of remaining interest, exchange offers, privatizations and clustered deals. 14 The target company is a U.S. or a non-u.s. firm which is either publicly held, private owned or a subsidiary of another company. The acquiring firm has sufficient coverage in the CRSP database. The ultimate parent of the acquiring firm is different from the ultimate parent of the target firm. The acquirer ownership of the target is less than or equal to 50% before the transaction and increases to greater than 50% after the transaction. Following the procedure used by Moeller, Schlingemann and Stulz (2004) and Billet and Qian (2008), the transaction value is at least $ 1 million and the deal value is at least 1% of the acquirer s pre-bid equity value, which is measured one month before the announcement date. Last, to prevent the results from being biased because of outliers I trim the acquirer cumulative abnormal return and the premium offered to the target. 15 The final acquisition history database consists of deals. 14 Clustered acquisitions are deals that are announced by the acquirer within 5 day from each other, thus it is difficult to isolate and attribute the acquirer s cumulative abnormal return for a particular clustered deal since it is affected by the announcement of another simultaneous deal by the same acquirer (Fuller et al., 2002). 15 I trim the acquirer cumulative abnormal returns at 1% and 99% levels following Alexandridis et al. (2012). The Premium paid to the Target (which is measured as the premium of the offer value to the Target's market value of equity 4 weeks prior to the announcement date, as reported by SDC) is trimmed beyond the standard acceptable range for premiums in empirical studies which is between [0,2] as suggested by Officer (2003) and Alexandridis et al. (2010). 9

10 The primary M&A sample that I examine consists of completed domestic public deals between 1995 and 2010 drawn from the initial M&A sample. 16 The final primary M&A sample consists of 2355 deals. To calculate the experience adjusted performance, the primary sample is split into a sub-sample of targets with acquisition experience (N = 1216) and a control sample of targets with no acquisition experience (N = 1339). 2.2 Sample Description Table 1 and Table 2 report the descriptive statistics of the sample by each year in the sample period and across different industries of the acquirer, respectively. 17 Table 1 highlights the general trend in M&A activity during the 1990s and the 2000s. The swell in the M&A activity can be seen with increase in the number of deals and also in the average transaction value (between ) during the 5 th merger wave of the 1990s and (between ) during the 6 th merger wave of 2000s as described by Alexandridis et. al. (2012). Table 2 indicates the clustering of M&A activity by industry as illustrated by Harford (2005). While Oil, Gas, and Coal Extraction and Products industry has the largest average transaction value of $ 5, million with 75 deals in the sample period, Consumer Durables industry had the lowest average transaction value of $ 1, million with only 29 deals in the sample period. I control for year and industry fixed effects in multivariate regression analysis. [Please insert Table 1 and Table 2 about here] Table 3 presents the descriptive statistics of the target's acquisition experience which is defined as the number of deals completed by the target in the prior 10 years or at least prior 5 years when the acquisition history of prior 10 years in not available. Using the above mentioned research design, the final acquisition sample consists of 2355 completed public deals, out of which 1139 (48.37%) deals involved a target with no acquisition experience and 16 The primary sample begins from 1995 to allow at least previous 5 years to measure the acquisition history of the acquirer and the target. 17 Industry is classified according to Fama-French 12 industry classification. 10

11 1216 (51.63%) deals involved a target acquisition experience of at least one deal. The procedure followed to calculate the experience adjusted performance measures for the subsample of target's with acquisition experience using the control-sample of targets with no acquisition experience is explained in detail in the following methodology section. For targets with acquisition experience, the mean (median) acquisition experience is 2.59 (2) deals. The sum of target's acquisition experience is 3151 deals and represents the acquisition activity of the targeted-acquirers before becoming a target. This further shows that public targets build significant level of acquisition experience before being acquired. [Please insert Table 3 about here] 2.3 Methodology and Variable Definitions I follow the standard event study methodology suggested by Brown and Warner (1985) to calculate the market reaction and measure the acquisition performance of the acquirer. It is measured by calculating the cumulative abnormal return (CAR) to the acquiring firm s shareholders over a 3-day (-1,+1) event window around the deal announcement date (ACAR). I estimate the abnormal returns using the market model, where CRSP value-weighted index is used as the market benchmark. 18 I calculate the experience adjusted acquisition performance variables (adjusted-acquirer CAR and adjusted-premium) to isolate the effect of target's acquisition experience on value creation. The experience adjusted performance allows me to examine the difference in the acquisition performance between acquiring a target with acquisition experience as compared to acquiring a similar target (i.e. with similar firm characteristics) but with no acquisition experience. The adjusted acquisition performance variables are defined as follows. 18 The market model parameters are estimated over a period of (-250,-15) around the deal announcement date, and the minimum estimation length is required to be equal to 30 days. 11

12 Adjusted acquirer CAR [Adjusted ACAR (-1,+1)] is calculated as acquirer CAR minus the median acquirer CAR of control sample of deals that were completed in the year around the announcement year and involved target in the same industry, same listing status, similar relative size (+/- 10%) and similar target MTB (+/- 10%). Adjusted-premium is calculated as premium minus the median premium of control sample of deals that were completed in the year around the announcement year and involved target in the same industry, same listing status, similar relative size (+/- 10%) and similar target MTB (+/- 10%). The premium is defined as the premium of the offer price to the target s share price 4 weeks prior to the deal announcement date (reported by SDC), with observations between zero and two. The control sample of deals includes acquisitions of targets with no acquisition experience. For matching the deals, target industry is classified according to Fama and French 12 industry classification. Since the target and the deal characteristics (including the announcement periods) are similar, the difference in acquisition performance can be attributed to the difference in the level of target's acquisition experience. This also allows adjusted-acquisition performance to control for the changes in performance due to the changes in firm's opportunity set suggested by Klasa and Stegemoller (2007). Moreover, Netter et al. (2012) mention that there is 3-fold drop in announcement returns between 1992 and Thus, if in general CARs have simply dropped over a period of time, then this will be controlled by calculating adjusted acquirer CAR. The main explanatory variables is target's acquisition experiences and it is defined as the number of deals completed by the target in the prior 10 years or at least prior 5 years when the acquisition history of prior 10 years in not available. I measure acquirer's acquisition experience using the definition consistent with the definition followed for target's acquisition experience. Although, the choice of 10 years is arbitrary, it provides sufficient time span to 12

13 allow the acquisition history to develop, but is adequate enough so that past acquisitions are likely to be informative. 19 That been said, I get similar results when I define acquisition experience based on 1) the entire sample period, and 2) the recent acquisition history built in the prior 5 years. 20 Variables related to the deal characteristics, the acquirer characteristics and the target characteristics are defined in the corresponding tables. Acquirer (and target) size and returns are calculated using information from CRSP. Accounting data is obtained from COMPUSTAT and all other deal characteristics are from SDC. 2.4 Deal and Firm Characteristics Table 4 reports deal and firm characteristics for the full sample and for different levels of target's acquisition experience. Panel A reports deal characteristics and Panel B reports predeal acquirer and target characteristics. I examine the difference in acquisition characteristics between targets with and without acquisition experience. Column 8 presents the difference in characteristics between acquisitions of targets with acquisition experience of at least 5 deals and those without any acquisition experience. 21 Similarly, column 9 presents the difference in acquisition characteristics between acquisitions of targets with acquisition experience of at least 5 deals and those with acquisition experience of 1 deal only. Next, to examine the direction and the strength of the linear association between a specific acquisition characteristic and the target's acquisition experience, I calculate and report the Pearson's 19 Although, it is practically difficult to disentangle the effect of firm learning and CEO learning, it is an area which demands further research. However it is important to note that firm learning is not inconsistent with CEO learning because firms can learn through CEOs. If there is no CEO turnover in the preceding years then firm acquisition experience will be the same as the CEO acquisition experience, which is more likely to be the case because CEO turnover is not common and therefore very small percentage of firms will be affected by CEO turnover events. Kaplan et al. (2012) reported that the rate of CEO Turnover from 1992 to 2007 is 15.8% with an average tenure as CEO of less than 7 years. Furthermore, given that the average CEO tenure is less than 10 years and my main results remain the same if I use 10-year, 5-year or the entire-sample pre-deal period to measure the acquisition history, suggests that the results are be robust to the CEO effect. 20 As an additional robustness check, I winsorize the target's and the acquirer's acquisition experience at 10 deals and re-run the tests. I find qualitatively similar results. 21 The difference tests are based on two-sample t-tests for means and Wilcoxon-sign rank tests for medians. ***,** and * denote statistical significance at the 1%, 5% and 10% level, respectively. 13

14 correlation coefficient between the given acquisition characteristic and the target's acquisition experience in column 10. [Please insert Table 4 about here] I find significant differences between acquisitions involving targets that develop acquisition experience as compared to acquisitions of targets that do not develop acquisition experience. Targets with acquisition experience and their acquirers are significantly larger in size. This is not surprising because firms grow as they acquire companies while it also takes larger firms to acquire more experienced (and hence larger) targets. This also reveals the intertwined nature of experience and size. While Schwert (2000) suggests that the bargaining power of the target firm depends on the firm size, Fuller, Netter and Stegemoller (2002) argue that larger the target relative to the acquirer, the stronger the target's negotiating position. However, Alexandridis et al. (2013) show that acquirers do not pay higher premiums but pay lower premiums for larger targets and still destroy value as they fail to deliver the expected synergy due to integration complexity involved in integrating large targets. Table 4 - Panel A shows that the relative size of the target to the acquirer is significantly larger for targets with acquisition experience as compared to those without any acquisition experience. This suggests that targets with acquisition experience hold a stronger position while negotiating the premium for their shareholders. To ensure that the experience effect that I document is beyond the size effect, I control for both acquirer and target size in cross-sectional regressions presented in the next section. 22 Acquisitions of targets with acquisition experience are less likely to have a pure cash payment or a pure stock payment, but more likely to have a mixed payment. If targets with 22 Another way to ensure that the target experience effect is free from the size effect is by examining the relation of premiums with experience. In their study, Alexandridis et al. (2013) show that acquirers pay lower premiums for larger targets due to high value-at-stake and deal complexity inherent in large deals. In contrast, I conjecture that targets with acquisition experience (and hence larger targets) will negotiate higher and not lower premiums because of the improvement in their ability to negotiate better deals as they learn with their experience from previous deals. I examine the predictions related to the premium received by the target in the following section and find evidence consistent with the above conjecture. 14

15 acquisition experience are larger targets than they are less likely to be financed with cash. (DeAngelo et al. 1984, Faccio and Massulis, 2005 and Alexandridis et al., 2013). Also, acquisitions driven by acquirer overvaluation are more likely to be paid for with stock (Shleifer and Vishny, 2003). The absence of significant relationship of pure cash and pure stock method of payment with target's acquisition experience suggests that under/overvaluation of experienced targets is less likely to motivate such acquisitions. Moreover, Travlos (1987) and Fuller et al. (2002) show that acquisitions of public targets paid with stock destroy value for the acquiring shareholders. I control for the effect of method of payment in multivariate analysis. The differences seen in column (8) and (9) suggest that deal characteristics are not significantly different between targets with acquisition experience as compared to targets without acquisition experience. This is further supported by the corresponding insignificant correlation coefficients seen in column (10), however, I find that deals during merger waves are significantly negatively correlated with target's acquisition experience, whereas deals with target termination fee are significantly positively correlated with target's acquisition experience. Mitchell and Lehn (1990) argue that value-destroying acquirers that get disciplined by the market are likely to resist their take-over and hence have a hostile attitude towards their acquirer's offer. Interestingly, results in Panel A highlight that target's attitude towards acquirer's offer does not become hostile or less friendly if the target has more acquisition experience. This suggests that targets with acquisition experience are not likely to be valuedestroying acquirers that get disciplined by the market for corporate control. Table 4 - Panel B shows that acquirers' Tobin's q, MTB and leverage (debt/assets) are not significantly different between acquisitions of targets with acquisition experience and those without any acquisition experience. If the firm's Tobin's q reflects its growth opportunities 15

16 (Billet and Qian, 2008) and management efficiency (Servaes, 1991 and Lang, Stulz and Walking, 1989) then we can conclude that acquirers of targets with acquisition experience do not have higher growth opportunities and do not employ better management teams relative to acquirers of targets without any acquisition experience. Moreover, the correlation coefficient of acquirer's FCF/assets with target's acquisition experience is significantly positive. Furthermore, although acquirer's acquisition experience increases with the target's acquisition experience, it does not happen at the same rate. On a closer look, the difference in acquirer's acquisition experience seen in column (8) (and column 9) suggests that when target's acquisition experience increases by 5 deals (4 deals), acquirer's acquisition experience on average increases by only 2.04 deals (1.89 deals). This suggests that targets with low level of acquisition experience interact with acquirers with high level or at the same level of acquisition experience, but targets with high level of acquisition experience interact with acquirers with low level of acquisition experience, which can give targets with high level of acquisition experience a stronger position while negotiating the deal. Lastly, I study the % of deals in which the acquirer remains independent till the end of the sample period (Non- Targeted Acquirer). In the full sample, 83.69% of deals involved non-targeted acquirers. When I classify the deals based on the target's acquisition experience, I find that for targets with no acquisition experience, 80.07% of deals involved non-targeted acquirers, but for targets with acquisition experience of 5 or more deals, 94.09% of deals involved non-targeted acquirer. The corresponding significantly positive correlation coefficient in column (10) suggests that the acquirers are more likely to remain independent after acquiring another experienced acquirer, and supports the defensive take-over argument suggested by Gorton, Kahl and Rosen (2009). 23 The characteristics of targets with acquisition experience are different from the characteristics of their acquirers. Statistics in Panel B show that target's debt/asset and FCF/assets are 23 A defensive acquirer may take-over other acquirers (experienced targets/ targeted acquirers) to reduce the threat of their own take-over. 16

17 significantly higher for targets with higher level of acquisition experience relative to targets without acquisition experience. Furthermore, the relation between target's Tobin's q and acquisition experience is significantly negative, which might hint that targets with acquisition experience face lower growth opportunities relative to targets without acquisition experience. The correlation coefficient between target's MTB ratio and acquisition experience is positive but insignificant, although the difference in median values of target's MTB for different levels of target's experience, seen in column (8) and (9) is positive and significant. Overall, the statistics of target characteristics in Panel B suggest that, targets with acquisition experience have higher debt/assets and FCF/assets, face limited growth opportunities and are likely to be valued higher relative to targets without acquisition experience. In next section, I present and discuss the empirical results. 3. Empirical Results In this section I investigate the relation of acquire CARs and adjusted acquirer CARs with target's acquisition experience. Table 5 presents the results from the univariate tests, and Table 6 presents the results from the multivariate tests. Overall, the evidence from univariate and multivariate tests shows that acquirer announcement returns decrease significantly with target's acquisition experience and is consistent with my hypothesis. 3.1 Univariate Analysis Table 5 reports the acquirer CARs and adjusted acquirer CARs around the announcement date classified by the acquisition experience of the target. The mean acquirer CAR for the full sample is -1.41%. When I classify the deals based on target's acquisition experience, I find that mean acquire CAR is -1.16% for targets with no acquisition experience but it is -2.95% for targets with acquisition experience of at least 5 deals. The difference of percentage points is statistically significant at 1% level. 17

18 [Please insert Table 5 about here] Furthermore, I report adjusted acquirer CAR calculated after using 2 different matching criteria. Adjusted ACAR1 is calculated when the sub-sample deals are matched with the control sample of deals based only on year of announcement and target's industry and listing status. 24 In adjusted ACAR2 I execute a more stringent matching criteria which is based on year of announcement, target's industry, listing status, similar relative size (+/-10%) and similar MTB ratio (+/-10%). 25 A more stringent matching criterion ensures that I compare highly similar deals while calculating adjusted acquirer CAR. Although I get similar results for adjusted ACAR1 and adjusted ACAR2, I focus on adjusted ACAR2 in the univariate and multivariate analysis since adjusted ACAR2 is calculated after using a stringent matching criterion. The mean adjusted acquirer CAR (adjusted ACAR2) for the sub-sample of targets with acquisition experience is -0.60%. The negative difference indicates that overall the acquirer CARs of deals that involve targets with acquisition experience are lower than the acquirer CARs of similar deals that involve similar targets with no acquisition experience. When I classify adjusted acquirer CAR based on target's acquisition experience, I find that adjusted acquirer CAR becomes more negative with increase in target's acquisition experience, suggesting that acquisition performance of acquirers that take-over targets with acquisition experience is worse than the acquisition performance of acquirers that take-over similar targets with no acquisition experience. The mean adjusted acquirer CAR is -0.03% for deals involving targets with acquisition experience of 1 deals, but it is -2.14% for deals involving 24 The sub-sample of deals consists of deals of targets with acquisition experience. The control sample of deals consists of deals of targets with no acquisition experience. 25 The full sample consists of 2355 deals consisting of 1139 deals of targets with no acquisition experience and 1216 deals of targets with acquisition experience. To calculate the adjusted acquirer CAR I match deals of targets with acquisition experience to deals of similar targets without acquisition experience. Therefore, when I analyze adjusted acquirer CAR the sub-sample size depends on the number of deals for which a match has been found. When I use a more stringent matching criterion the sub-sample size reduces because it is difficult to find the right match for some deals due to the strong restrictions imposed on the matching procedure. Hence, a stringent matching criterion ensures comparison of highly similar deals but reduces the sub-sample size. 18

19 targets with acquisition experience of at least 5 deal. The difference of percentage points is significant at 1% level. Figure 1 graphically illustrates the relation of acquirer s acquisition performance (acquirer CAR and adjusted acquirer CAR) with target s acquisition experience. The reduction in acquirer's gains with increase in target's acquisition experience is evident from the graph. [Please insert Figure 1 about here] In summary, the results from the univariate analysis suggest that acquirer CARs and adjusted acquirer CARs decline with increase in the targets acquisition experience which is consistent with my hypothesis, and support the conjecture that experienced targets learn from their past deals and negotiate better deals which leads to the reduction in the extent of gains captured by the acquirer. However, in the above discussed comparisons I ignored that firms and deals differ in other dimensions, which could be driving the results. I take these determinants of acquirer CARs into account in the multivariate analysis and present it in the next section. 3.2 Multivariate Analysis Table 6 presents the results of ordinary least squares regression of the acquirer s 3-day cumulative abnormal return [ACAR (-1,+1)] and adjusted cumulative abnormal return [ACAR (-1,+1)] on the target s acquisition experience and other control variables. 26 In this multivariate framework I control for acquirer's acquisition experience, deal and firm characteristics along with year and industry effects. In all the specifications, I find evidence consistent with my hypothesis that predicts significant negative relation between acquirer gains (acquirer CAR and adjusted acquirer CAR) and the target's acquisition experience. 27 The coefficient estimates of the target's 26 To ensure robustness, I also calculate abnormal returns using the modified market model and after altering the event window to (-2,+2). The results remain qualitatively unaltered from the main results. 27 Performing the regression analysis based on White-adjusted standard errors produces qualitatively similar results. 19

20 acquisition experience are negative and statistically significant in all the specifications. 28 In specification (3), just 1 deal increase in the target's acquisition experience reduces the gains to the acquiring shareholders by 0.16 percentage points, a reduction of 11.35% given the mean acquirer CAR of -1.41%. If we consider the mean value of the acquirer's market capitalisation ($ 10, mil) this translates into an average drop in value of $ 17.4 million for the acquiring shareholders over a 3-day period around the announcement date. Furthermore, columns (8) and (9) indicate significant negative relation between adjusted acquirer CARs and target's acquisition experience, which suggests that the acquisition performance of acquirers that take-over target's with acquisition experience is worse than the acquisition performance of acquirers that take-over similar target's with no acquisition experience. [Please insert Table 6 about here] The coefficient estimate of the acquirer's acquisition experience is statistically insignificant, indicating that the acquirer's acquisition experience does not affect acquirer gains. Acquirers unable to learn with experience (because of hubris) predicts significant negative relation between acquirer gains and acquirer's acquisition experience (Billet and Qian, 2008), whereas learning predicts the opposite. However, considering that public deals mostly destroy value (Fuller, Netter and Stegemoller, 2002), no value destruction with acquirer's acquisition experience could be interpreted as value preservation as the acquirer gains experience. I also examined for a U-shaped relation between the acquirer gains and the acquirer's (as well as the 28 To ensure that the target experience effect documented in this study is beyond the size effect noted in previous studies it is essential to control for both acquirer and target size, however there is a possibility that the correlations between these variables could affect the results, especially because the correlation between target size and acquirer size is 50.9%. To verify that my results are not affected by model specifications, I run the following ACAR regressions 1) only with target's experience, 2) with target size, acquirer size and relative size, each separately and 3) with target size, acquirer size and relative size, altogether. In all the specifications, I find qualitatively similar results; the coefficient of the target's acquisition experience is significantly negative at 5% level. Previous studies that included both target and acquirer size in their abnormal-returns model specification include Officer (2003) and Baker et al., (2012). 20

21 target's) acquisition experience, but did not find any evidence to support that. The coefficient estimates of control variable are consistent with the finding of previous papers. Since more experienced targets are larger and tend to be acquired by larger bidder, it is important to control for acquirer and target size in the cross-sectional regressions. Acquirer CARs are significantly negatively related to the size of the target. Alexandridis et al. (2013) find that acquirer announcement returns are significantly lower for larger targets even though they are paid lower premiums by the acquirer after anticipating the integration complexity associated with combining a larger target. Although target's with acquisition experience are more likely to be larger targets, I do not expect them to accept lower premium if they learn to negotiate effectively from their previous deals. I examine premiums in the next section and find that targets with acquisition experience do not accept lower premiums but negotiate higher premiums for their shareholders. To completely rule out the possibility that target size is driving the results to any extent, I re-run the regressions for small targets only (i.e. market cap. below the median target in our sample). The coefficient of target s acquisition experience (= ) remains significant at 5% level indicating that the negative association between acquirer CARs and target experience persists among deals involving smaller targets. Furthermore, I find that all stock public deals have lower acquirer CARs, consistent with Travlos (1987) and Fuller, Netter and Stegemoller (2002). However, similar to Schwert (2000) I do not find that larger acquirers experience lower CARs as suggested by Moeller, Schlingemann and Stulz (2004). Moreover only in specifications (1) and (2), I find weak evidence that acquirer tobin's q is negatively related to acquirer CAR in line with Moeller, Schlingemann and Stulz (2004, 2005), but inconsistent with Servaes (1991) who finds that for public acquisition high q bidders have higher abnormal returns. For other specifications I find insignificant relation between acquirer Tobin's q and acquirer abnormal returns. Other control variables such as diversifying deal dummy, hostile deal dummy, competed deal dummy and target's tobin's q have insignificant coefficients. 21

22 Overall, after controlling for deal and firm characteristics along with year and industry fixed effects, I find that acquirer gains decline with increase in the target's acquisition experience. In the next section, I examine the driver behind the reduction in acquirer gains with the target's acquisition experience. 4. Premium and Target's Acquisition Experience So far, we have seen that acquirer gains reduce with increase in the targets acquisition experience. In this section, I investigate whether the source of this reduction in acquirer gains is related to the higher premiums negotiated by experienced targets. If targets negotiate higher premiums because of their acquisition experience then that could explain the reduction in acquirer CARs with increase in the target's acquisition experience. Certainly, as targets learn from their previous acquisitions and draw inferences from their past deals, it can enable them to precisely evaluate the synergies associated with the acquirers (Aktas et al., 2009), and hence set an effective reserve price to negotiate higher premiums (Fuller et al., 2002) for their shareholders. Therefore, I examine the relation between the premium and the target s acquisition experience after controlling for other factors that affect premium. Table 7 reports the results of ordinary least squares regression of the premium and adjustedpremium on the target s acquisition experience and other control variables. Premium is defined as the premium of the offer price to the target s share price 4 weeks prior to the deal announcement date, with observations between zero and two. 29 Adjusted-premium is the difference in the premium received by a target with acquisition experience and the median value of premium received by similar targets with no acquisition experience around the same time period. Similar to adjusted acquirer CARs, adjusted premium is more likely to isolate 29 The values of premium are reported by SDC. For robustness, I also examine other alternate definitions of premiums such as a) the Schwert s premium defined as the cumulative abnormal return to the target shareholders for the (-63,126) window around the deal announcement date following Schwert (2000) and b) the premium of offer price reported by SDC over the target s share price one month prior to the deal announcement date obtained from CRSP. I find qualitatively similar results. 22